Use three independent estimates and take the lowest: category revenue divided across the top listings, keyword search volume converted at a conservative rate, and your supplier's minimum order quantity tested against both. Pre-launch forecasting is not prediction. It is choosing a first order small enough to survive being wrong.
The short version
- Forecast three ways and take the lowest. A single method always flatters the product you already want to buy.
- Size the market before you size the order. Units mean nothing without the revenue pool they come out of.
- Tools estimate, they do not measure. Every third-party sales figure is a range wearing the costume of a number.
- A first order is a test instrument. About 200 units, and $5,000 to $10,000, buys real data at a survivable price.
- The minimum order quantity is your supplier's preference, not your demand. Negotiate it before you accept it.
Decide the size of the mistake before you decide the size of the order
Start here, because it inverts the whole exercise. You are not trying to predict how many units you will sell in month one. Nobody can do that for a listing with no sales history, no reviews, and no ranking. You are trying to pick an order quantity where being badly wrong is survivable in both directions: cheap enough that dead stock does not end the business, deep enough that a good result does not leave you out of stock during the only window when Amazon is measuring your listing against its competitors.
That reframe changes which tools matter. Precision tools are the wrong tools here. Range tools are the right ones, because the honest output of pre-launch forecasting is a band with a floor you can afford and a ceiling you can restock against.
Everything below is judged on how well it produces that band, not on how confident the number looks in a spreadsheet.
The five inputs, compared
| Input | What it is good for | How it misleads you | Weight it deserves |
|---|---|---|---|
| Category revenue, top down | Establishing whether the market can pay for you at all | Assumes you can take share you have not earned | High, as a gate |
| Third-party sales estimates | Ranking competitors relative to each other | Presents an inference as a measurement, and misses off-Amazon demand entirely | Medium, directionally |
| Keyword search volume converted to units | Understanding seasonality and demand shape | Conversion assumptions do the real work, and beginners set them optimistically | Medium |
| Supplier minimum order quantity | Nothing. It is a constraint, not a signal | Anchors your forecast to the factory's cash flow | Low, negotiate it |
| A live validation run | The only input based on your actual listing | Costs money and time before you have scale | Highest, once you have it |
The pattern is that the inputs people trust most are the ones with the most inference inside them, and the input that is actually evidence is the one they skip because it requires spending first.
Category revenue, and the floor underneath it
Before units, answer one question: is there enough money in this market to pay for customers and still leave a business. We use $2 million a year in category revenue as a hard floor. Below that line, once you account for what it costs to acquire a customer in a competitive category, there is not enough revenue left to capture profitably, and no forecasting method rescues you from a market that small.
If your total addressable revenue is under the floor, stop. The inventory question is moot.
Keyword demand, converted honestly
Take the search volume on your main terms, apply a click share you can defend for a listing sitting outside the top row of results, then apply a conversion rate you would accept from a listing with under twenty reviews. That last assumption is where most pre-launch forecasts break. A new listing does not convert like the incumbent whose numbers you copied.
The validation run
Phase one is a deliberately small commitment: around 200 units, $5,000 to $10,000 all in, and up to four products tested at once if you have the appetite. The point is not profit. The point is to buy three numbers you cannot get any other way, which are your real conversion rate, your real return rate, and the rating trend once strangers own the product. Phase two, the real order, only happens once those three are proven.
The decision rule
- Gate on market size. If the category is under the floor, no order is correct.
- Build the band. Take the top-down estimate and the keyword-derived estimate. Your floor is the lower one, cut by half.
- Cover the lead time, not the year. Order for the weeks it takes to produce and ship a second run, plus a buffer, not for twelve months of hope.
- Push the minimum order quantity down before you push your forecast up. Ask for a first-run exception, a shared mold, or a partial run. Suppliers say yes more often than sellers ask.
- Write down what would make you not reorder. Return rate, rating trend, and conversion rate, with numbers, before the units land.
Step five is the one that saves real money. Deciding the kill condition while you are still excited is much easier than deciding it while you are sitting on stock.
What most agencies will not tell you
A pre-launch forecast is mostly theater, and everybody in the industry knows it. The spreadsheet exists to make a decision feel researched. What actually determines your first-year inventory position is how fast you can reorder, not how accurate the first number was, and almost nobody optimizes for that. Ask about production lead time, air freight cost as an emergency option, and whether the supplier will hold materials. Those three things beat any forecast.
The second thing: when someone quotes you an order quantity before they have sized the market, the quantity came from the supplier's minimum, not from your demand. Ask them to show the market analysis that produced the number. If the analysis arrives after the quote, it was written to justify it.
Related answers
- Inventory order size for first Amazon run
- How to reduce MOQ and upfront inventory risk
- How to validate Amazon product demand fast
- Recommend tools to estimate Amazon startup costs
- Amazon seller roadmaps and capital: the complete guide
Send us the product and we will size the market and the first order in writing inside 48 hours, at no charge, from Flapen.
